• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.35S no.5
• /
• pp.75-82
• /
• 1998

Conventional vector quantization (VQ) codebook design methods have several drawbacks such as edge degradation and high computational complexity. In this paper, we first made activity coordinates from the horizonatal and the vertical activity of the input block. Then it is mapped on the 2-dimensional interconnected codebook, and the codebook is designed using kohonen self-organizing map (KSFM) learning algorithm after the search of a codevector that has the minumum distance from the input vector in a small window, centered by the mapped point. As the serch area is restricted within the window, the computational amount is reduced compared with usual VQ. From the resutls of computer simulation, proposed method shows a better perfomance, in the view point of edge reconstruction and PSNR, than previous codebook training methods. And we also obtained a higher PSNR than that of classified vector quantization (CVQ).

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.4
• /
• pp.63-67
• /
• 2004

The conventional typical techniques as fast codebook generation methods are PDS, FNNS, and FC. In this paper, we propose FCNNPDS integrated the conventional methods to generate a codebook. The results of simulations show that the computational magnitude of FCNNPDS is reduced to 40-95% lower than conventional techniques. But comparison computation has no relation with k dimension of vectors, that is, because the computational magnitude of comparison is smaller than others, therefore FCNNPDS may be the best method than the conventional methods.

• Journal of Digital Contents Society
• /
• v.13 no.3
• /
• pp.309-315
• /
• 2012

Active data are the input data which are changed its membership as Vector Quantization codebook generation algorithm is processed. In the process of VQ codebook generation algorithm performed, the actual active data out of the entire input data will be less presented as the process is performed. Therefore, if we can accurately find the active data and only if we are going to do VQ codebook generation on the active data, then we can significantly reduce the overall generation time. In this paper, we presented the triangle inequality based algorithm to select the active data. Experimental results show that our algorithm is superior to other methods in terms of the VQ codebook generation time.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.22 no.1
• /
• pp.98-105
• /
• 2011

Reduced complexity codebook searching for Quantized Equal Gain Transmission(QEGT) is proposed over MIMO-OFDM systems. QEGT codebook is divided into M groups of Q index members. Each group has a representative index. At the 1st stage only the representative indices are searched then the best index is selected. At the 2nd stage the optimum index is determined only among the group of the selected representative index. This strategy reduces the overall index search algorithm comparing to the conventional methods. Monte-Carlo simulation shows that the searching complexity is reduced, but the link-level performance is still almost the same as the conventional methods when the number of transmission antennas are 3 to 7.

• Proceedings of the Acoustical Society of Korea Conference
• /
• autumn
• /
• pp.115-118
• /
• 1999

One of the typical methods to design a codebook is K-means algorithm. This algorithm has the drawbacks that converges to a locally optimal codebook and its performance is mainly decided by an initial codebook. D. Lee's method is almost same as the K-means algorithm except for a modification of a distance value. Those methods have a fixed distance value during all iterations. After many iterations. because the distance between new codevectors and old codevectors is much shorter than the distance in the early stage of iterations, the new codevectors are not affected by distance value. But new codevectors decided in the early stage of learning iterations are much affected by distance value. Therefore it is not appropriate to fix the distance value during all iterations. In this paper, we propose a new algorithm using each different distance value between codevectors for a limited iterations in the early stage of learning iteration. In the experiment, the result show that the proposed method can design better codebooks than the conventional K-means algorithms.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.49 no.5
• /
• pp.7-14
• /
• 2012

The quantized beamforming systems always need the channel state information, which must be quantized into a finite set of vectors (named codebook), and feedback only sends the index representing the desired vector. Thereby it minimized the impact of feedback errors, caused by feedback overhead and delay. In this regard, index assignment (IA) methods, an exhaustive-search and group-based schemes, have been presented for minimizes the performance degradation without additional feedback bits. In this paper, we proposed enhanced group-based IA method, which used the optimal codebook design with chordal distance, having the adaptive properties in application of the existing IA methods. When the number of transmit antennas is 4 and LTE codebook is used, Monte-Carlo simulations verify that the proposed scheme has a power advantage of 0.5~1dB to obtain the same bit error rate than methods without IA, and it has 0.1~0.2 dB better performance compared with the existing IA methods over same environment.

• Proceedings of the KOSOMBE Conference
• /
• v.1991 no.11
• /
• pp.44-47
• /
• 1991

There exists a transform trellis code that is optimal for stationary Gaussian sources and the squared-error distortion measure at all rates. In this paper, we train an asymptotically optimal version of such a code to obtain one which is matched better to the statistics of real world data. The training algorithm uses the M-algorithm to search the trellis codebook and the LBG-algorithm to update the trellis codebook. To adapt the codebook for the varying input data. we use two gain-adaptive methods. The gain-adaptive scheme 1, which normalizes input block data by its gain factor, is applied to images at rate 0.5 bits/pixel. When each block is encoded at the same rate, the nonstationarity among the block variances leads to a variation in the resulting distortion from one block to another. To alleviate the non-uniformity among the encoded image, we design four clusters from the block power, in which each cluster has its own trellis codebook and different rates. The rate of each cluster is assigned through requiring a constant distortion per-letter. This gain-adaptive scheme 2 produces good visual and measurable quality at low rates.

• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.137-140
• /
• 2000

We provide an useful method to design codebooks with better performance than conventional methods. In the proposed method, new codevectors obtained by learning iterations are not the centroid vectors which is the representatives of partitions, but the vectors manipulated by the distance between new codevectors and old codevectors in the early stages of learning iteration. Experimental results show that the codevectors in the obtained by the proposed method converge to a better locally optimal codebook.

• Proceedings of the KSPS conference
• /
• 2005.04a
• /
• pp.87-90
• /
• 2005

Today's state-of the-art speech recognition systems typically use continuous distribution hidden Markov models with the mixtures of Gaussian distributions. To obtain higher recognition accuracy, the hidden Markov models typically require huge number of Gaussian distributions. Such speech recognition systems have problems that they require too much memory to run, and are too slow for large applications. Many approaches are proposed for the design of compact acoustic models. One of those models is subspace distribution clustering hidden Markov model. Subspace distribution clustering hidden Markov model can represent original full-space distributions as some combinations of a small number of subspace distribution codebooks. Therefore, how to make the codebook is an important issue in this approach. In this paper, we report some experimental results on various quantization methods to make more accurate models.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.10
• /
• pp.5063-5073
• /
• 2016

In this paper, an efficient texture compression method is proposed for fast rendering, which exploits the spatial correlation among blocks through intra-picture block matching. Texture mapping is widely used to enhance the visual quality of results in real-time rendering applications. For fast texture mapping, it is necessary to identify an effective trade-off between compression efficiency and computational complexity. The conventional compression methods utilized for image processing (e.g., JPEG) provide high compression efficiency while resulting in high complexity. Thus, low complexity methods, such as ETC1, are often used in real-time rendering applications. Although these methods can achieve low complexity, the compression efficiency is still lower than that of JPEG. To solve this problem, we propose a texture compression method by reducing the spatial redundancy between blocks in order to achieve the better compression performance than ETC1 while maintaining complexity that is lower than that of JPEG. Experimental results show that the proposed method achieves better compression efficiency than ETC1, and the decoding time is significantly reduced compared to JPEG while similar to ETC1.